[PATCH 0/1] Add uvirtio for testing

[PATCH 0/1] Add uvirtio for testing

[Lepton Wu]

This is a way to create virtio based devices from user space. This is the
background for this patch:

We have some images works fine under qemu, we'd like to also run the same image
on Google Cloud. Currently Google Cloud doesn't support virtio-vga. I had a 
patch to create a virtio-vga from kernel directly:
https://www.spinics.net/lists/dri-devel/msg248573.html

Then I got feedback from Gerd that maybe it's better to change that to something
like uvirtio. Since I really don't have other use cases for now, I just implemented the minimal stuff which work for my use case.

Lepton Wu (1):
  virtio: Add uvirtio driver

 drivers/virtio/Kconfig        |   8 +
 drivers/virtio/Makefile       |   1 +
 drivers/virtio/uvirtio.c      | 405 ++++++++++++++++++++++++++++++++++
 include/linux/uvirtio.h       |   8 +
 include/uapi/linux/uvirtio.h  |  69 ++++++
 samples/uvirtio/Makefile      |   9 +
 samples/uvirtio/uvirtio-vga.c |  63 ++++++
 7 files changed, 563 insertions(+)
 create mode 100644 drivers/virtio/uvirtio.c
 create mode 100644 include/linux/uvirtio.h
 create mode 100644 include/uapi/linux/uvirtio.h
 create mode 100644 samples/uvirtio/Makefile
 create mode 100644 samples/uvirtio/uvirtio-vga.c

-- 
2.26.2.303.gf8c07b1a785-goog

[PATCH 1/1] virtio: Add uvirtio driver

[Lepton Wu]

This is for testing purpose to create virtio devices from user space.
uvirtio-vga.c shows how to create a virtio-vga device. Currently we
don't have a use case which requires user/kernel communication so
read/write api hasn't been implemented.

Signed-off-by: Lepton Wu <ytht.net@gmail.com>
---
 drivers/virtio/Kconfig        |   8 +
 drivers/virtio/Makefile       |   1 +
 drivers/virtio/uvirtio.c      | 405 ++++++++++++++++++++++++++++++++++
 include/linux/uvirtio.h       |   8 +
 include/uapi/linux/uvirtio.h  |  69 ++++++
 samples/uvirtio/Makefile      |   9 +
 samples/uvirtio/uvirtio-vga.c |  63 ++++++
 7 files changed, 563 insertions(+)
 create mode 100644 drivers/virtio/uvirtio.c
 create mode 100644 include/linux/uvirtio.h
 create mode 100644 include/uapi/linux/uvirtio.h
 create mode 100644 samples/uvirtio/Makefile
 create mode 100644 samples/uvirtio/uvirtio-vga.c

diff --git a/drivers/virtio/Kconfig b/drivers/virtio/Kconfig
index 69a32dfc318a..8bfb77db4efa 100644
--- a/drivers/virtio/Kconfig
+++ b/drivers/virtio/Kconfig
@@ -109,4 +109,12 @@ config VIRTIO_MMIO_CMDLINE_DEVICES
 
 	 If unsure, say 'N'.
 
+config UVIRTIO
+	tristate "UVirtio driver"
+	select VIRTIO
+	---help---
+	 This driver supports create virtio devices from guest userspace.
+
+	 If unsure, say 'N'.
+
 endif # VIRTIO_MENU
diff --git a/drivers/virtio/Makefile b/drivers/virtio/Makefile
index 29a1386ecc03..558b2f890e8c 100644
--- a/drivers/virtio/Makefile
+++ b/drivers/virtio/Makefile
@@ -7,3 +7,4 @@ virtio_pci-$(CONFIG_VIRTIO_PCI_LEGACY) += virtio_pci_legacy.o
 obj-$(CONFIG_VIRTIO_BALLOON) += virtio_balloon.o
 obj-$(CONFIG_VIRTIO_INPUT) += virtio_input.o
 obj-$(CONFIG_VIRTIO_VDPA) += virtio_vdpa.o
+obj-$(CONFIG_UVIRTIO) += uvirtio.o
diff --git a/drivers/virtio/uvirtio.c b/drivers/virtio/uvirtio.c
new file mode 100644
index 000000000000..a25107473dab
--- /dev/null
+++ b/drivers/virtio/uvirtio.c
@@ -0,0 +1,405 @@
+/*
+ *  User level driver support for virtio subsystem
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ */
+
+#include <linux/module.h>
+#include <linux/miscdevice.h>
+#include <linux/slab.h>
+#include <linux/uaccess.h>
+#include <linux/uvirtio.h>
+#include <linux/virtio.h>
+#include <linux/virtio_config.h>
+#include <linux/virtio_ring.h>
+#include <uapi/linux/uvirtio.h>
+
+#define UVIRTIO_MAX_EXPECT_DATA  (1UL << 20)
+
+struct uvirtio_device {
+	struct virtio_device vdev;
+	struct mutex mutex;
+	enum uvirtio_state state;
+	unsigned char virtio_status;
+	struct uvirtio_setup setup;
+	struct uvirtio_expect expect;
+	char *expect_data;
+};
+
+static struct miscdevice uvirtio_misc;
+
+static struct bus_type uvirtio_bus = {
+	.name = "",
+};
+
+static u64 uvirtio_get_features(struct virtio_device *dev)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	return udev->setup.features;
+}
+
+static int uvirtio_finalize_features(struct virtio_device *vdev)
+{
+	return 0;
+}
+
+static void uvirtio_get(struct virtio_device *dev, unsigned int offset,
+			void *buf, unsigned int len)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	if (WARN_ON(offset + len > udev->setup.config_len))
+		return;
+	memcpy(buf, (char *)udev->setup.config_addr + offset, len);
+}
+
+static u8 uvirtio_get_status(struct virtio_device *dev)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	return udev->virtio_status;
+}
+
+static void uvirtio_set_status(struct virtio_device *dev, u8 status)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	if (WARN_ON(!status))
+		return;
+	udev->virtio_status = status;
+}
+
+static int find_match(int write, char *buf, unsigned int len,
+		      struct uvirtio_block *block, char *data)
+{
+	int i;
+	int off = 0;
+	for (i = 0; i < UVIRTIO_MAX_RULES; ++i) {
+		if (!block->rules[i].len)
+			break;
+		if (block->rules[i].off + block->rules[i].len > len)
+			return -1;
+		if (write) {
+			memcpy(buf + block->rules[i].off,
+			       data + block->data + off, block->rules[i].len);
+		} else {
+			if (memcmp(buf + block->rules[i].off,
+				   data + block->data + off,
+				   block->rules[i].len))
+				return -1;
+		}
+		off += block->rules[i].len;
+	}
+	return i ? write : 0;
+}
+
+static void process_vq(struct virtio_device *vdev, const char *name,
+		       struct vring_desc *desc, int idx)
+{
+	unsigned short flags;
+	char *buf;
+	unsigned int len;
+	struct uvirtio_device *udev = container_of(vdev, struct uvirtio_device,
+						   vdev);
+	struct uvirtio_expect *expect = &udev->expect;
+	int i, j, matched = 0;
+
+	for (i = 0; i < UVIRTIO_MAX_EXPECTS; ++i) {
+		if (!expect->expects[i].vq_name[0])
+			break;
+		if (strncmp(name, expect->expects[i].vq_name,
+			    sizeof(expect->expects[i].vq_name)))
+			continue;
+		for (j = 0; j < UVIRTIO_MAX_BLOCKS &&
+		     expect->expects[i].blocks[j].len; ++j) {
+			flags = virtio16_to_cpu(vdev, desc[idx].flags);
+			len = virtio32_to_cpu(vdev, desc[idx].len);
+			buf = __va(virtio64_to_cpu(vdev, desc[idx].addr));
+			if (expect->expects[i].blocks[j].len != len ||
+			    expect->expects[i].blocks[j].flags != flags)
+				break;
+			matched = find_match(flags & VRING_DESC_F_WRITE,
+					     buf, len,
+					     &expect->expects[i].blocks[j],
+					     udev->expect_data);
+			if (matched)
+				break;
+			if (!(flags & VRING_DESC_F_NEXT))
+				break;
+			idx = virtio16_to_cpu(vdev, desc[idx].next);
+		}
+		if (matched > 0)
+			break;
+	}
+}
+
+static bool uvirtio_notify(struct virtqueue *vq)
+{
+	struct vring *r = (struct vring *)(vq + 1);
+	int used_idx, avail_idx, id;
+	used_idx = virtio16_to_cpu(vq->vdev, r->used->idx);
+	avail_idx = virtio16_to_cpu(vq->vdev, r->avail->idx);
+	while (used_idx != avail_idx) {
+		id = used_idx & (r->num - 1);
+		process_vq(vq->vdev, vq->name, r->desc, r->avail->ring[id]);
+		r->used->ring[id].id = r->avail->ring[id];
+		used_idx++;
+	}
+	r->used->idx = r->avail->idx;
+	vq->callback(vq);
+	return true;
+}
+
+static int uvirtio_find_vqs(struct virtio_device *dev, unsigned int nvqs,
+			    struct virtqueue *vqs[],
+			    vq_callback_t * callbacks[],
+			    const char *const names[],
+			    const bool *ctx, struct irq_affinity *desc)
+{
+	int i, j;
+
+	for (i = 0; i < nvqs; ++i) {
+		vqs[i] = vring_create_virtqueue(i, 256, SMP_CACHE_BYTES, dev,
+						true, false, false,
+						uvirtio_notify, callbacks[i],
+						names[i]);
+		if (!vqs[i])
+			goto err;
+	}
+	return 0;
+err:
+	for (j = 0; j < i; ++j) {
+		vring_del_virtqueue(vqs[j]);
+		vqs[j] = NULL;
+	}
+	return -ENOMEM;
+}
+
+static void uvirtio_del_vqs(struct virtio_device *dev)
+{
+	struct virtqueue *vq, *n;
+
+	list_for_each_entry_safe(vq, n, &dev->vqs, list)
+		vring_del_virtqueue(vq);
+}
+
+static void uvirtio_reset(struct virtio_device *vdev)
+{
+}
+
+static const struct virtio_config_ops uvirtio_vq_ops = {
+	.get_features = uvirtio_get_features,
+	.finalize_features = uvirtio_finalize_features,
+	.get = uvirtio_get,
+	.get_status = uvirtio_get_status,
+	.set_status = uvirtio_set_status,
+	.reset = uvirtio_reset,
+	.find_vqs = uvirtio_find_vqs,
+	.del_vqs = uvirtio_del_vqs,
+};
+
+static int uvirtio_open(struct inode *inode, struct file *file)
+{
+	struct uvirtio_device *newdev;
+
+	newdev = kzalloc(sizeof(*newdev), GFP_KERNEL);
+	if (!newdev)
+		return -ENOMEM;
+
+	mutex_init(&newdev->mutex);
+	newdev->state = UVST_NEW_DEVICE;
+
+	file->private_data = newdev;
+	nonseekable_open(inode, file);
+
+	return 0;
+}
+
+static int uvirtio_release(struct inode *inode, struct file *file)
+{
+	struct uvirtio_device *udev = file->private_data;
+
+	if (udev->state == UVST_CREATED) {
+		if (udev->setup.flags & (1 << UVIRTIO_DEV_FLAG_STICK))
+			return 0;
+		unregister_virtio_device(&udev->vdev);
+	}
+	kfree((void *)udev->setup.config_addr);
+	kfree(udev->expect_data);
+	kfree(udev);
+	return 0;
+}
+
+static int load_data(struct uvirtio_block *b, char **ptr,
+		     unsigned int *off, unsigned int *len)
+{
+	int i, ret, total = 0;
+	void __user *p = (void __user *)b->data;
+	unsigned int need;
+	int realloc = 0;
+
+	for (i = 0; i < UVIRTIO_MAX_RULES; ++i) {
+		if (!b->rules[i].len)
+			break;
+		total += b->rules[i].len;
+	}
+	if (!total) {
+		return 0;
+	}
+
+	need = (*off) + total;
+	while (need > (*len)) {
+		if (*len >= UVIRTIO_MAX_EXPECT_DATA)
+			return -ENOMEM;
+		*len = (*len) << 1;
+		realloc = 1;
+	}
+	if (realloc)
+		(*ptr) = krealloc(*ptr, *len, GFP_KERNEL);
+	if (!(*ptr))
+		return -ENOMEM;
+
+	ret = copy_from_user(*ptr + (*off), p, total);
+	if (ret)
+		return ret;
+	b->data = (*off);
+	(*off) += total;
+	return 0;
+}
+
+static int set_expect(struct uvirtio_device *udev, unsigned long arg)
+{
+	int i, j, ret;
+	struct uvirtio_block *b;
+	unsigned int off = 0, len = 1024;
+
+	ret = copy_from_user(&udev->expect, (void __user *)arg,
+			     sizeof(udev->expect));
+	if (ret)
+		return ret;
+
+	udev->expect_data = kmalloc(len, GFP_KERNEL);
+
+	if (!udev->expect_data)
+		return -ENOMEM;
+
+	for (i = 0; i < UVIRTIO_MAX_EXPECTS; ++i) {
+		if (!udev->expect.expects[i].vq_name[0])
+			break;
+		for (j = 0; j < UVIRTIO_MAX_BLOCKS; ++j) {
+			b = &udev->expect.expects[i].blocks[j];
+			if (!b->len)
+				break;
+			ret = load_data(b, &udev->expect_data, &off, &len);
+			if (ret)
+				goto err;
+		}
+	}
+	return 0;
+err:
+	kfree(udev->expect_data);
+	udev->expect_data = NULL;
+	return ret;
+}
+
+static int create_device(struct uvirtio_device *udev, unsigned long arg)
+{
+	int ret;
+	char *config = NULL;
+	ret = copy_from_user(&udev->setup, (void __user *)arg,
+			     sizeof(udev->setup));
+	if (ret)
+		return ret;
+	if (!udev->setup.config_len || !udev->setup.config_addr)
+		return -EINVAL;
+	if (udev->setup.config_len > UVIRTIO_MAX_CONFIG_LEN)
+		return -EINVAL;
+	if (udev->setup.flags & ~((1 << UVIRTIO_DEV_FLAG_MAX) - 1))
+		return -EINVAL;
+	config = kmalloc(udev->setup.config_len, GFP_KERNEL);
+	if (!config)
+		return -ENOMEM;
+	ret = copy_from_user(config, (void __user *)udev->setup.config_addr,
+			     udev->setup.config_len);
+	if (ret)
+		goto err;
+	udev->setup.config_addr = (uint64_t) config;
+	if (uvirtio_misc.this_device->bus == NULL) {
+		uvirtio_misc.this_device->bus = &uvirtio_bus;
+	}
+	udev->vdev.dev.parent = uvirtio_misc.this_device;
+	udev->vdev.id.device = udev->setup.id;
+	udev->vdev.config = &uvirtio_vq_ops;
+	ret = register_virtio_device(&udev->vdev);
+	if (ret)
+		goto err;
+	udev->state = UVST_CREATED;
+	return 0;
+err:
+	kfree(config);
+	udev->setup.config_addr = 0;
+	return ret;
+}
+
+static long uvirtio_ioctl(struct file *file, unsigned int cmd,
+			  unsigned long arg)
+{
+	struct uvirtio_device *udev = file->private_data;
+	int ret;
+
+	ret = mutex_lock_interruptible(&udev->mutex);
+	if (ret)
+		return ret;
+
+	if (udev->state != UVST_NEW_DEVICE) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	switch (cmd) {
+	case UV_DEV_CREATE:
+		ret = create_device(udev, arg);
+		break;
+	case UV_DEV_EXPECT:
+		ret = set_expect(udev, arg);
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+out:
+	mutex_unlock(&udev->mutex);
+	return ret;
+}
+
+static const struct file_operations uvirtio_fops = {
+	.owner = THIS_MODULE,
+	.open = uvirtio_open,
+	.release = uvirtio_release,
+	.unlocked_ioctl = uvirtio_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl = uvirtio_ioctl,
+#endif
+	.llseek = no_llseek,
+};
+
+static struct miscdevice uvirtio_misc = {
+	.fops = &uvirtio_fops,
+	.minor = MISC_DYNAMIC_MINOR,
+	.name = "uvirtio",
+};
+
+module_misc_device(uvirtio_misc);
+
+MODULE_VERSION("0.0.1");
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("User level driver support for virtio subsystem");
diff --git a/include/linux/uvirtio.h b/include/linux/uvirtio.h
new file mode 100644
index 000000000000..ff217080d921
--- /dev/null
+++ b/include/linux/uvirtio.h
@@ -0,0 +1,8 @@
+#ifndef __UVIRTIO_H_
+#define __UVIRTIO_H_
+
+#include <uapi/linux/uvirtio.h>
+
+enum uvirtio_state { UVST_NEW_DEVICE, UVST_CREATED };
+
+#endif  /* __UVIRTIO_H_ */
diff --git a/include/uapi/linux/uvirtio.h b/include/uapi/linux/uvirtio.h
new file mode 100644
index 000000000000..4771e527fce0
--- /dev/null
+++ b/include/uapi/linux/uvirtio.h
@@ -0,0 +1,69 @@
+/* SPDX-License-Identifier: GPL-2.0+ WITH Linux-syscall-note */
+/*
+ *  User level driver support for virtio subsystem
+ *
+ * Based on uinput.c by Aristeu Sergio Rozanski Filho
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _UAPI_UVIRTIO_H_
+#define _UAPI_UVIRTIO_H_
+
+#include <linux/types.h>
+
+/* ioctl */
+
+#define UVIRTIO_MAX_CONFIG_LEN 1024
+#define UVIRTIO_MAX_RULES 8
+#define UVIRTIO_MAX_BLOCKS 8
+#define UVIRTIO_MAX_EXPECTS 8
+#define UVIRTIO_MAX_VQ_NAME 64
+
+enum {
+	UVIRTIO_DEV_FLAG_STICK = 0,
+	UVIRTIO_DEV_FLAG_MAX
+};
+
+struct uvirtio_setup {
+	__u64 features;
+	__u64 config_addr;
+	__u16 config_len;
+	__u16 flags;
+	__u32 id;
+};
+
+struct uvirtio_block {
+	__u32 len;
+	__u16 flags;
+	__u16 unused;
+	__u64 data;
+	struct {
+		__u16 off;
+		__u16 len;
+	} rules[UVIRTIO_MAX_RULES];
+};
+
+struct uvirtio_expect {
+	struct {
+		char vq_name[UVIRTIO_MAX_VQ_NAME];
+		struct uvirtio_block blocks[UVIRTIO_MAX_BLOCKS];
+	} expects[UVIRTIO_MAX_EXPECTS];
+};
+
+#define UVIRTIO_IOCTL_BASE	'V'
+#define UV_DEV_CREATE		_IOW(UVIRTIO_IOCTL_BASE, 1, struct uvirtio_setup)
+#define UV_DEV_EXPECT		_IOW(UVIRTIO_IOCTL_BASE, 2, struct uvirtio_expect)
+
+#endif /* _UAPI_UVIRTIO_H_ */
diff --git a/samples/uvirtio/Makefile b/samples/uvirtio/Makefile
new file mode 100644
index 000000000000..ee830986eda2
--- /dev/null
+++ b/samples/uvirtio/Makefile
@@ -0,0 +1,9 @@
+uvirtio-vga: uvirtio-vga.c uapi/linux/uvirtio.h linux/uvirtio.h
+	$(CC) -o $@ $^ -I.
+uapi/linux/uvirtio.h:
+	mkdir -p uapi/linux && ln ../../../../include/uapi/linux/uvirtio.h uapi/linux -sf
+linux/uvirtio.h:
+	mkdir -p linux && ln ../../../include/linux/uvirtio.h linux -sf
+.PHONY: clean
+clean:
+	rm -f uvirtio-vga
diff --git a/samples/uvirtio/uvirtio-vga.c b/samples/uvirtio/uvirtio-vga.c
new file mode 100644
index 000000000000..b65cb62a6d91
--- /dev/null
+++ b/samples/uvirtio/uvirtio-vga.c
@@ -0,0 +1,63 @@
+#include <stdio.h>
+#include <stddef.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/ioctl.h>
+#include <linux/uvirtio.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <linux/virtio_config.h>
+#include <linux/virtio_ids.h>
+#include <linux/virtio_gpu.h>
+#include <linux/virtio_ring.h>
+
+void error(char *str)
+{
+	fprintf(stderr, "%s\n", str);
+	exit(-1);
+}
+
+int main()
+{
+	int fd = open("/dev/uvirtio", O_RDWR);
+	if (fd < 0) {
+		error("open");
+	}
+	struct uvirtio_expect expect = { };
+
+	strcpy(expect.expects[0].vq_name, "control");
+	struct uvirtio_block *b = &expect.expects[0].blocks[0];
+	struct virtio_gpu_ctrl_hdr cmd;
+	cmd.type = VIRTIO_GPU_CMD_GET_DISPLAY_INFO;
+	b->len = sizeof(cmd);
+	b->flags = VRING_DESC_F_NEXT;
+	b->data = (__u64) & cmd.type;
+	b->rules[0].off = offsetof(struct virtio_gpu_ctrl_hdr, type);
+	b->rules[0].len = sizeof(cmd.type);
+	struct virtio_gpu_resp_display_info info = { };
+	info.pmodes[0].r.width = 1024;
+	info.pmodes[0].r.height = 768;
+	info.pmodes[0].enabled = 1;
+	b = &expect.expects[0].blocks[1];
+	b->len = sizeof(info);
+	b->flags = VRING_DESC_F_WRITE;
+	b->data = (__u64) & info.pmodes[0];
+	b->rules[0].off =
+	    offsetof(struct virtio_gpu_resp_display_info, pmodes[0]);
+	b->rules[0].len = sizeof(info.pmodes[0]);
+	int ret = ioctl(fd, UV_DEV_EXPECT, &expect);
+	if (ret < 0)
+		error("ioctl UV_DEV_EXPECT");
+	struct uvirtio_setup setup;
+	struct virtio_gpu_config config = {.num_scanouts = 1 };
+	setup.features = 1ULL << VIRTIO_F_VERSION_1;
+	setup.config_addr = (__u64) & config;
+	setup.config_len = sizeof(config);
+	setup.id = VIRTIO_ID_GPU;
+	setup.flags = 1 << UVIRTIO_DEV_FLAG_STICK;
+	ret = ioctl(fd, UV_DEV_CREATE, &setup);
+	if (ret < 0)
+		error("ioctl UV_DEV_CREATE");
+}
-- 
2.26.2.303.gf8c07b1a785-goog

Re: [PATCH 0/1] Add uvirtio for testing

[Jason Wang]

On 2020/4/29 上午4:47, Lepton Wu wrote:
> This is a way to create virtio based devices from user space. This is the
> background for this patch:
>
> We have some images works fine under qemu, we'd like to also run the same image
> on Google Cloud. Currently Google Cloud doesn't support virtio-vga. I had a
> patch to create a virtio-vga from kernel directly:
> https://www.spinics.net/lists/dri-devel/msg248573.html
>
> Then I got feedback from Gerd that maybe it's better to change that to something
> like uvirtio. Since I really don't have other use cases for now, I just implemented the minimal stuff which work for my use case.


Interesting, several questions:

1) Are you aware of virtio vhost-user driver done by UM guys? 
(arch/um/drivers/virtio_uml.c) The memory part is tricky but overall 
both of you have similar target.
2) Patch 1 said it's userspace virtio driver, which I think it is 
actually "userspace virtio device"
3) Need to be verbose on how the vring processing work in the commit log 
of patch 1
4) I'm curious which testing you want to accomplish through this new 
transport, I guess you want to test a specific virtio driver?
5) You mentioned that you may want to develop communication between 
kernel and userspace, any more details on that?

Thanks


>
> Lepton Wu (1):
>    virtio: Add uvirtio driver
>
>   drivers/virtio/Kconfig        |   8 +
>   drivers/virtio/Makefile       |   1 +
>   drivers/virtio/uvirtio.c      | 405 ++++++++++++++++++++++++++++++++++
>   include/linux/uvirtio.h       |   8 +
>   include/uapi/linux/uvirtio.h  |  69 ++++++
>   samples/uvirtio/Makefile      |   9 +
>   samples/uvirtio/uvirtio-vga.c |  63 ++++++
>   7 files changed, 563 insertions(+)
>   create mode 100644 drivers/virtio/uvirtio.c
>   create mode 100644 include/linux/uvirtio.h
>   create mode 100644 include/uapi/linux/uvirtio.h
>   create mode 100644 samples/uvirtio/Makefile
>   create mode 100644 samples/uvirtio/uvirtio-vga.c
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH 0/1] Add uvirtio for testing

[Gerd Hoffmann]

> 3) Need to be verbose on how the vring processing work in the commit log of
> patch 1

Ecven better a file documenting the interface somewhere in
Documentation/

take care,
  Gerd

[PATCH v2] virtio: Add uvirtio driver

[Lepton Wu]

This is for testing purpose to create virtio devices from user space.
uvirtio-vga.c shows how to create a virtio-vga device.

For "simple" device like virtio 2d VGA, actually we only need to handle
one virtio request which return the resolution of VGA. We provide a
UV_DEV_EXPECT ioctl which set the expected virtio request and the prepared
reply. This can eliminate user/kernel communication. We just handle this in
the notify callback of the virtqueue.

Check samples/uvirtio/uvirtio-vga.c for example.

Currently we don't have a use case which requires user/kernel communication so
read/write api hasn't been implemented.

Signed-off-by: Lepton Wu <ytht.net@gmail.com>
---
v2:
  * Fix styles issues found by checkpatch.pl
  * Update comments and commit log
---
 drivers/virtio/Kconfig        |  11 +
 drivers/virtio/Makefile       |   1 +
 drivers/virtio/uvirtio.c      | 399 ++++++++++++++++++++++++++++++++++
 include/linux/uvirtio.h       |  16 ++
 include/uapi/linux/uvirtio.h  |  50 +++++
 samples/uvirtio/Makefile      |   9 +
 samples/uvirtio/uvirtio-vga.c |  72 ++++++
 7 files changed, 558 insertions(+)
 create mode 100644 drivers/virtio/uvirtio.c
 create mode 100644 include/linux/uvirtio.h
 create mode 100644 include/uapi/linux/uvirtio.h
 create mode 100644 samples/uvirtio/Makefile
 create mode 100644 samples/uvirtio/uvirtio-vga.c

diff --git a/drivers/virtio/Kconfig b/drivers/virtio/Kconfig
index 69a32dfc318a..4686df49cac5 100644
--- a/drivers/virtio/Kconfig
+++ b/drivers/virtio/Kconfig
@@ -109,4 +109,15 @@ config VIRTIO_MMIO_CMDLINE_DEVICES
 
 	 If unsure, say 'N'.
 
+config UVIRTIO
+	tristate "UVirtio driver"
+	select VIRTIO
+	help
+	 This driver supports creating virtio devices from userspace.
+
+	 This can be used to create virtio devices from user space without
+         supports from VMM. Check samples/uvirtio for examples.
+
+	 If unsure, say 'N'.
+
 endif # VIRTIO_MENU
diff --git a/drivers/virtio/Makefile b/drivers/virtio/Makefile
index 29a1386ecc03..558b2f890e8c 100644
--- a/drivers/virtio/Makefile
+++ b/drivers/virtio/Makefile
@@ -7,3 +7,4 @@ virtio_pci-$(CONFIG_VIRTIO_PCI_LEGACY) += virtio_pci_legacy.o
 obj-$(CONFIG_VIRTIO_BALLOON) += virtio_balloon.o
 obj-$(CONFIG_VIRTIO_INPUT) += virtio_input.o
 obj-$(CONFIG_VIRTIO_VDPA) += virtio_vdpa.o
+obj-$(CONFIG_UVIRTIO) += uvirtio.o
diff --git a/drivers/virtio/uvirtio.c b/drivers/virtio/uvirtio.c
new file mode 100644
index 000000000000..64cc9140de7a
--- /dev/null
+++ b/drivers/virtio/uvirtio.c
@@ -0,0 +1,399 @@
+// SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note
+/*
+ *  User level device support for virtio subsystem
+ */
+
+#include <linux/module.h>
+#include <linux/miscdevice.h>
+#include <linux/slab.h>
+#include <linux/uaccess.h>
+#include <linux/uvirtio.h>
+#include <linux/virtio.h>
+#include <linux/virtio_config.h>
+#include <linux/virtio_ring.h>
+#include <uapi/linux/uvirtio.h>
+
+#define UVIRTIO_MAX_EXPECT_DATA  (1UL << 20)
+
+struct uvirtio_device {
+	struct virtio_device vdev;
+	struct mutex mutex;
+	enum uvirtio_state state;
+	unsigned char virtio_status;
+	struct uvirtio_setup setup;
+	struct uvirtio_expect expect;
+	char *expect_data;
+};
+
+static struct miscdevice uvirtio_misc;
+
+static struct bus_type uvirtio_bus = {
+	.name = "",
+};
+
+static u64 uvirtio_get_features(struct virtio_device *dev)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	return udev->setup.features;
+}
+
+static int uvirtio_finalize_features(struct virtio_device *vdev)
+{
+	return 0;
+}
+
+static void uvirtio_get(struct virtio_device *dev, unsigned int offset,
+			void *buf, unsigned int len)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	if (WARN_ON(offset + len > udev->setup.config_len))
+		return;
+	memcpy(buf, (char *)udev->setup.config_addr + offset, len);
+}
+
+static u8 uvirtio_get_status(struct virtio_device *dev)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	return udev->virtio_status;
+}
+
+static void uvirtio_set_status(struct virtio_device *dev, u8 status)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	if (WARN_ON(!status))
+		return;
+	udev->virtio_status = status;
+}
+
+static int find_match(int write, char *buf, unsigned int len,
+		      struct uvirtio_block *block, char *data)
+{
+	int i;
+	int off = 0;
+
+	for (i = 0; i < UVIRTIO_MAX_RULES; ++i) {
+		if (!block->rules[i].len)
+			break;
+		if (block->rules[i].off + block->rules[i].len > len)
+			return -1;
+		if (write) {
+			memcpy(buf + block->rules[i].off,
+			       data + block->data + off, block->rules[i].len);
+		} else {
+			if (memcmp(buf + block->rules[i].off,
+				   data + block->data + off,
+				   block->rules[i].len))
+				return -1;
+		}
+		off += block->rules[i].len;
+	}
+	return i ? write : 0;
+}
+
+static void process_vq(struct virtio_device *vdev, const char *name,
+		       struct vring_desc *desc, int idx)
+{
+	unsigned short flags;
+	char *buf;
+	unsigned int len;
+	struct uvirtio_device *udev = container_of(vdev, struct uvirtio_device,
+						   vdev);
+	struct uvirtio_expect *expect = &udev->expect;
+	int i, j, matched = 0;
+
+	for (i = 0; i < UVIRTIO_MAX_EXPECTS; ++i) {
+		if (!expect->expects[i].vq_name[0])
+			break;
+		if (strncmp(name, expect->expects[i].vq_name,
+			    sizeof(expect->expects[i].vq_name)))
+			continue;
+		for (j = 0; j < UVIRTIO_MAX_BLOCKS &&
+		     expect->expects[i].blocks[j].len; ++j) {
+			flags = virtio16_to_cpu(vdev, desc[idx].flags);
+			len = virtio32_to_cpu(vdev, desc[idx].len);
+			buf = __va(virtio64_to_cpu(vdev, desc[idx].addr));
+			if (expect->expects[i].blocks[j].len != len ||
+			    expect->expects[i].blocks[j].flags != flags)
+				break;
+			matched = find_match(flags & VRING_DESC_F_WRITE,
+					     buf, len,
+					     &expect->expects[i].blocks[j],
+					     udev->expect_data);
+			if (matched)
+				break;
+			if (!(flags & VRING_DESC_F_NEXT))
+				break;
+			idx = virtio16_to_cpu(vdev, desc[idx].next);
+		}
+		if (matched > 0)
+			break;
+	}
+}
+
+static bool uvirtio_notify(struct virtqueue *vq)
+{
+	struct vring *r = (struct vring *)(vq + 1);
+	int used_idx, avail_idx, id;
+
+	used_idx = virtio16_to_cpu(vq->vdev, r->used->idx);
+	avail_idx = virtio16_to_cpu(vq->vdev, r->avail->idx);
+	while (used_idx != avail_idx) {
+		id = used_idx & (r->num - 1);
+		process_vq(vq->vdev, vq->name, r->desc, r->avail->ring[id]);
+		r->used->ring[id].id = r->avail->ring[id];
+		used_idx++;
+	}
+	r->used->idx = r->avail->idx;
+	vq->callback(vq);
+	return true;
+}
+
+static int uvirtio_find_vqs(struct virtio_device *dev, unsigned int nvqs,
+			    struct virtqueue *vqs[],
+			    vq_callback_t *callbacks[],
+			    const char *const names[],
+			    const bool *ctx, struct irq_affinity *desc)
+{
+	int i, j;
+
+	for (i = 0; i < nvqs; ++i) {
+		vqs[i] = vring_create_virtqueue(i, 256, SMP_CACHE_BYTES, dev,
+						true, false, false,
+						uvirtio_notify, callbacks[i],
+						names[i]);
+		if (!vqs[i])
+			goto err;
+	}
+	return 0;
+err:
+	for (j = 0; j < i; ++j) {
+		vring_del_virtqueue(vqs[j]);
+		vqs[j] = NULL;
+	}
+	return -ENOMEM;
+}
+
+static void uvirtio_del_vqs(struct virtio_device *dev)
+{
+	struct virtqueue *vq, *n;
+
+	list_for_each_entry_safe(vq, n, &dev->vqs, list)
+		vring_del_virtqueue(vq);
+}
+
+static void uvirtio_reset(struct virtio_device *vdev)
+{
+}
+
+static const struct virtio_config_ops uvirtio_vq_ops = {
+	.get_features = uvirtio_get_features,
+	.finalize_features = uvirtio_finalize_features,
+	.get = uvirtio_get,
+	.get_status = uvirtio_get_status,
+	.set_status = uvirtio_set_status,
+	.reset = uvirtio_reset,
+	.find_vqs = uvirtio_find_vqs,
+	.del_vqs = uvirtio_del_vqs,
+};
+
+static int uvirtio_open(struct inode *inode, struct file *file)
+{
+	struct uvirtio_device *newdev;
+
+	newdev = kzalloc(sizeof(*newdev), GFP_KERNEL);
+	if (!newdev)
+		return -ENOMEM;
+
+	mutex_init(&newdev->mutex);
+	newdev->state = UVST_NEW_DEVICE;
+
+	file->private_data = newdev;
+	nonseekable_open(inode, file);
+
+	return 0;
+}
+
+static int uvirtio_release(struct inode *inode, struct file *file)
+{
+	struct uvirtio_device *udev = file->private_data;
+
+	if (udev->state == UVST_CREATED) {
+		if (udev->setup.flags & (1 << UVIRTIO_DEV_FLAG_STICK))
+			return 0;
+		unregister_virtio_device(&udev->vdev);
+	}
+	kfree((void *)udev->setup.config_addr);
+	kfree(udev->expect_data);
+	kfree(udev);
+	return 0;
+}
+
+static int load_data(struct uvirtio_block *b, char **ptr,
+		     unsigned int *off, unsigned int *len)
+{
+	int i, ret, total = 0;
+	void __user *p = (void __user *)b->data;
+	unsigned int need;
+	int realloc = 0;
+
+	for (i = 0; i < UVIRTIO_MAX_RULES; ++i) {
+		if (!b->rules[i].len)
+			break;
+		total += b->rules[i].len;
+	}
+	if (!total)
+		return 0;
+
+	need = (*off) + total;
+	while (need > (*len)) {
+		if (*len >= UVIRTIO_MAX_EXPECT_DATA)
+			return -ENOMEM;
+		*len = (*len) << 1;
+		realloc = 1;
+	}
+	if (realloc)
+		(*ptr) = krealloc(*ptr, *len, GFP_KERNEL);
+	if (!(*ptr))
+		return -ENOMEM;
+
+	ret = copy_from_user(*ptr + (*off), p, total);
+	if (ret)
+		return ret;
+	b->data = (*off);
+	(*off) += total;
+	return 0;
+}
+
+static int set_expect(struct uvirtio_device *udev, unsigned long arg)
+{
+	int i, j, ret;
+	struct uvirtio_block *b;
+	unsigned int off = 0, len = 1024;
+
+	ret = copy_from_user(&udev->expect, (void __user *)arg,
+			     sizeof(udev->expect));
+	if (ret)
+		return ret;
+
+	udev->expect_data = kmalloc(len, GFP_KERNEL);
+
+	if (!udev->expect_data)
+		return -ENOMEM;
+
+	for (i = 0; i < UVIRTIO_MAX_EXPECTS; ++i) {
+		if (!udev->expect.expects[i].vq_name[0])
+			break;
+		for (j = 0; j < UVIRTIO_MAX_BLOCKS; ++j) {
+			b = &udev->expect.expects[i].blocks[j];
+			if (!b->len)
+				break;
+			ret = load_data(b, &udev->expect_data, &off, &len);
+			if (ret)
+				goto err;
+		}
+	}
+	return 0;
+err:
+	kfree(udev->expect_data);
+	udev->expect_data = NULL;
+	return ret;
+}
+
+static int create_device(struct uvirtio_device *udev, unsigned long arg)
+{
+	int ret;
+	char *config = NULL;
+
+	ret = copy_from_user(&udev->setup, (void __user *)arg,
+			     sizeof(udev->setup));
+	if (ret)
+		return ret;
+	if (!udev->setup.config_len || !udev->setup.config_addr)
+		return -EINVAL;
+	if (udev->setup.config_len > UVIRTIO_MAX_CONFIG_LEN)
+		return -EINVAL;
+	if (udev->setup.flags & ~((1 << UVIRTIO_DEV_FLAG_MAX) - 1))
+		return -EINVAL;
+	config = kmalloc(udev->setup.config_len, GFP_KERNEL);
+	if (!config)
+		return -ENOMEM;
+	ret = copy_from_user(config, (void __user *)udev->setup.config_addr,
+			     udev->setup.config_len);
+	if (ret)
+		goto err;
+	udev->setup.config_addr = (uint64_t) config;
+
+	if (uvirtio_misc.this_device->bus == NULL)
+		uvirtio_misc.this_device->bus = &uvirtio_bus;
+
+	udev->vdev.dev.parent = uvirtio_misc.this_device;
+	udev->vdev.id.device = udev->setup.id;
+	udev->vdev.config = &uvirtio_vq_ops;
+	ret = register_virtio_device(&udev->vdev);
+	if (ret)
+		goto err;
+	udev->state = UVST_CREATED;
+	return 0;
+err:
+	kfree(config);
+	udev->setup.config_addr = 0;
+	return ret;
+}
+
+static long uvirtio_ioctl(struct file *file, unsigned int cmd,
+			  unsigned long arg)
+{
+	struct uvirtio_device *udev = file->private_data;
+	int ret;
+
+	ret = mutex_lock_interruptible(&udev->mutex);
+	if (ret)
+		return ret;
+
+	if (udev->state != UVST_NEW_DEVICE) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	switch (cmd) {
+	case UV_DEV_CREATE:
+		ret = create_device(udev, arg);
+		break;
+	case UV_DEV_EXPECT:
+		ret = set_expect(udev, arg);
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+out:
+	mutex_unlock(&udev->mutex);
+	return ret;
+}
+
+static const struct file_operations uvirtio_fops = {
+	.owner = THIS_MODULE,
+	.open = uvirtio_open,
+	.release = uvirtio_release,
+	.unlocked_ioctl = uvirtio_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl = uvirtio_ioctl,
+#endif
+	.llseek = no_llseek,
+};
+
+static struct miscdevice uvirtio_misc = {
+	.fops = &uvirtio_fops,
+	.minor = MISC_DYNAMIC_MINOR,
+	.name = "uvirtio",
+};
+
+module_misc_device(uvirtio_misc);
+
+MODULE_VERSION("0.0.1");
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("User level driver support for virtio subsystem");
diff --git a/include/linux/uvirtio.h b/include/linux/uvirtio.h
new file mode 100644
index 000000000000..ae3fd6dfbe58
--- /dev/null
+++ b/include/linux/uvirtio.h
@@ -0,0 +1,16 @@
+/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
+/*
+ * User level device support for virtio subsystem
+ *
+ * Check samples/uvirtio for examples.
+ *
+ * Based on uinput.c by Aristeu Sergio Rozanski Filho
+ */
+#ifndef __UVIRTIO_H_
+#define __UVIRTIO_H_
+
+#include <uapi/linux/uvirtio.h>
+
+enum uvirtio_state { UVST_NEW_DEVICE, UVST_CREATED };
+
+#endif  /* __UVIRTIO_H_ */
diff --git a/include/uapi/linux/uvirtio.h b/include/uapi/linux/uvirtio.h
new file mode 100644
index 000000000000..a17f923d73a8
--- /dev/null
+++ b/include/uapi/linux/uvirtio.h
@@ -0,0 +1,50 @@
+/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
+#ifndef _UAPI_UVIRTIO_H_
+#define _UAPI_UVIRTIO_H_
+
+#include <linux/types.h>
+
+/* ioctl */
+
+#define UVIRTIO_MAX_CONFIG_LEN 1024
+#define UVIRTIO_MAX_RULES 8
+#define UVIRTIO_MAX_BLOCKS 8
+#define UVIRTIO_MAX_EXPECTS 8
+#define UVIRTIO_MAX_VQ_NAME 64
+
+enum {
+	UVIRTIO_DEV_FLAG_STICK = 0,
+	UVIRTIO_DEV_FLAG_MAX
+};
+
+struct uvirtio_setup {
+	__u64 features;
+	__u64 config_addr;
+	__u16 config_len;
+	__u16 flags;
+	__u32 id;
+};
+
+struct uvirtio_block {
+	__u32 len;
+	__u16 flags;
+	__u16 unused;
+	__u64 data;
+	struct {
+		__u16 off;
+		__u16 len;
+	} rules[UVIRTIO_MAX_RULES];
+};
+
+struct uvirtio_expect {
+	struct {
+		char vq_name[UVIRTIO_MAX_VQ_NAME];
+		struct uvirtio_block blocks[UVIRTIO_MAX_BLOCKS];
+	} expects[UVIRTIO_MAX_EXPECTS];
+};
+
+#define UVIRTIO_IOCTL_BASE	'V'
+#define UV_DEV_CREATE	_IOW(UVIRTIO_IOCTL_BASE, 1, struct uvirtio_setup)
+#define UV_DEV_EXPECT	_IOW(UVIRTIO_IOCTL_BASE, 2, struct uvirtio_expect)
+
+#endif /* _UAPI_UVIRTIO_H_ */
diff --git a/samples/uvirtio/Makefile b/samples/uvirtio/Makefile
new file mode 100644
index 000000000000..ee830986eda2
--- /dev/null
+++ b/samples/uvirtio/Makefile
@@ -0,0 +1,9 @@
+uvirtio-vga: uvirtio-vga.c uapi/linux/uvirtio.h linux/uvirtio.h
+	$(CC) -o $@ $^ -I.
+uapi/linux/uvirtio.h:
+	mkdir -p uapi/linux && ln ../../../../include/uapi/linux/uvirtio.h uapi/linux -sf
+linux/uvirtio.h:
+	mkdir -p linux && ln ../../../include/linux/uvirtio.h linux -sf
+.PHONY: clean
+clean:
+	rm -f uvirtio-vga
diff --git a/samples/uvirtio/uvirtio-vga.c b/samples/uvirtio/uvirtio-vga.c
new file mode 100644
index 000000000000..fae79a6f4968
--- /dev/null
+++ b/samples/uvirtio/uvirtio-vga.c
@@ -0,0 +1,72 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <stdio.h>
+#include <stddef.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/ioctl.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <linux/uvirtio.h>
+#include <linux/virtio_config.h>
+#include <linux/virtio_ids.h>
+#include <linux/virtio_gpu.h>
+#include <linux/virtio_ring.h>
+
+void error(char *str)
+{
+	fprintf(stderr, "%s\n", str);
+	exit(-1);
+}
+
+int main(void)
+{
+	int ret;
+	int fd = open("/dev/uvirtio", O_RDWR);
+
+	if (fd < 0)
+		error("open");
+
+	struct uvirtio_expect expect = { };
+
+	strcpy(expect.expects[0].vq_name, "control");
+
+	struct uvirtio_block *b = &expect.expects[0].blocks[0];
+	struct virtio_gpu_ctrl_hdr cmd;
+
+	cmd.type = VIRTIO_GPU_CMD_GET_DISPLAY_INFO;
+	b->len = sizeof(cmd);
+	b->flags = VRING_DESC_F_NEXT;
+	b->data = (__u64) &cmd.type;
+	b->rules[0].off = offsetof(struct virtio_gpu_ctrl_hdr, type);
+	b->rules[0].len = sizeof(cmd.type);
+
+	struct virtio_gpu_resp_display_info info = { };
+
+	info.pmodes[0].r.width = 1024;
+	info.pmodes[0].r.height = 768;
+	info.pmodes[0].enabled = 1;
+	b = &expect.expects[0].blocks[1];
+	b->len = sizeof(info);
+	b->flags = VRING_DESC_F_WRITE;
+	b->data = (__u64) &info.pmodes[0];
+	b->rules[0].off =
+	    offsetof(struct virtio_gpu_resp_display_info, pmodes[0]);
+	b->rules[0].len = sizeof(info.pmodes[0]);
+	ret = ioctl(fd, UV_DEV_EXPECT, &expect);
+	if (ret < 0)
+		error("ioctl UV_DEV_EXPECT");
+
+	struct uvirtio_setup setup;
+	struct virtio_gpu_config config = {.num_scanouts = 1 };
+
+	setup.features = 1ULL << VIRTIO_F_VERSION_1;
+	setup.config_addr = (__u64) &config;
+	setup.config_len = sizeof(config);
+	setup.id = VIRTIO_ID_GPU;
+	setup.flags = 1 << UVIRTIO_DEV_FLAG_STICK;
+	ret = ioctl(fd, UV_DEV_CREATE, &setup);
+	if (ret < 0)
+		error("ioctl UV_DEV_CREATE");
+}
-- 
2.26.2.303.gf8c07b1a785-goog

Re: [PATCH 0/1] Add uvirtio for testing

[lepton]

On Wed, Apr 29, 2020 at 2:57 AM Jason Wang <jasowang@redhat.com> wrote:
>
>
> On 2020/4/29 上午4:47, Lepton Wu wrote:
> > This is a way to create virtio based devices from user space. This is the
> > background for this patch:
> >
> > We have some images works fine under qemu, we'd like to also run the same image
> > on Google Cloud. Currently Google Cloud doesn't support virtio-vga. I had a
> > patch to create a virtio-vga from kernel directly:
> > https://www.spinics.net/lists/dri-devel/msg248573.html
> >
> > Then I got feedback from Gerd that maybe it's better to change that to something
> > like uvirtio. Since I really don't have other use cases for now, I just implemented the minimal stuff which work for my use case.
>
>
> Interesting, several questions:
>
> 1) Are you aware of virtio vhost-user driver done by UM guys?
> (arch/um/drivers/virtio_uml.c) The memory part is tricky but overall
> both of you have similar target.
Thanks for reminding me, I was not aware of it. The use case looks a
little different: they are trying create virtio devices for user mode
linux and it communicated with "host" side. My driver doesn't depends
on any HOST/VMM side stuff. Basically we can use it to create virtio
device from guest itself. Or even create virtio device on bare metal.
> 2) Patch 1 said it's userspace virtio driver, which I think it is
> actually "userspace virtio device"
Updated in version 2 of this patch.
> 3) Need to be verbose on how the vring processing work in the commit log
> of patch 1
Updated.
> 4) I'm curious which testing you want to accomplish through this new
> transport, I guess you want to test a specific virtio driver?
Here is the whole story: we want to test our custom linux image. In
the past, we just test our custom linux image with qemu (and virtio
vga), and run qemu session on Google Cloud. As you can see, this is
nested virtualization and performance hurts. And more, we have another
vm inside our custom linux image. So we want to remove the qemu layer,
just run our custom linux image on Google Cloud directly. Then we need
some kind of VGA.  So a "dummy" virtio vga looks a good fit.
> 5) You mentioned that you may want to develop communication between
> kernel and userspace, any more details on that?
Currently, we don't have such use case. But maybe others can
furthermore to extend uvirtio for this. For example, user space can
use read/write to actually exchange data with kernel. Then that could
be used for simulate more complex use case.
>
> Thanks
>
>
> >
> > Lepton Wu (1):
> >    virtio: Add uvirtio driver
> >
> >   drivers/virtio/Kconfig        |   8 +
> >   drivers/virtio/Makefile       |   1 +
> >   drivers/virtio/uvirtio.c      | 405 ++++++++++++++++++++++++++++++++++
> >   include/linux/uvirtio.h       |   8 +
> >   include/uapi/linux/uvirtio.h  |  69 ++++++
> >   samples/uvirtio/Makefile      |   9 +
> >   samples/uvirtio/uvirtio-vga.c |  63 ++++++
> >   7 files changed, 563 insertions(+)
> >   create mode 100644 drivers/virtio/uvirtio.c
> >   create mode 100644 include/linux/uvirtio.h
> >   create mode 100644 include/uapi/linux/uvirtio.h
> >   create mode 100644 samples/uvirtio/Makefile
> >   create mode 100644 samples/uvirtio/uvirtio-vga.c
> >
>
_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH 0/1] Add uvirtio for testing

[lepton]

On Wed, Apr 29, 2020 at 4:58 AM Gerd Hoffmann <kraxel@redhat.com> wrote:
>
> > 3) Need to be verbose on how the vring processing work in the commit log of
> > patch 1
>
> Ecven better a file documenting the interface somewhere in
> Documentation/
I put a uvirtio-vga.c under samples/uvirtio and hope this can serve
the purpose of the document since currently that's the only tested use
case. Maybe have a document later if this really get more use cases?
Thanks.
>
> take care,
>   Gerd
>

Re: [PATCH 0/1] Add uvirtio for testing

[Gerd Hoffmann]

On Wed, Apr 29, 2020 at 08:59:18PM -0700, lepton wrote:
> On Wed, Apr 29, 2020 at 4:58 AM Gerd Hoffmann <kraxel@redhat.com> wrote:
> >
> > > 3) Need to be verbose on how the vring processing work in the commit log of
> > > patch 1
> >
> > Ecven better a file documenting the interface somewhere in
> > Documentation/
> I put a uvirtio-vga.c under samples/uvirtio and hope this can serve
> the purpose of the document since currently that's the only tested use
> case.

Sample code is good, but *commented* sample code would be better.

> Maybe have a document later if this really get more use cases?

Well, having good documentation right from the start would help getting
more use cases ;)

take care,
  Gerd

[PATCH v3] virtio: Add uvirtio driver

[Lepton Wu]

This is for testing purpose to create virtio devices from user space.
uvirtio-vga.c shows how to create a virtio-vga device.

For "simple" device like virtio 2d VGA, actually we only need to handle
one virtio request which return the resolution of VGA. We provide a
UV_DEV_EXPECT ioctl which set the expected virtio request and the prepared
reply. This can eliminate user/kernel communication. We just handle this in
the notify callback of the virtqueue.

Check samples/uvirtio/uvirtio-vga.c for example.

Currently we don't have a use case which requires user/kernel communication
so read/write API hasn't been implemented.

Signed-off-by: Lepton Wu <ytht.net@gmail.com>
---
v3:
  * Add Documentation/misc-devices/uvirtio.rst
v2:
  * Fix styles issues found by checkpatch.pl
  * Update comments and commit log
---
 Documentation/misc-devices/index.rst   |   1 +
 Documentation/misc-devices/uvirtio.rst |  56 ++++
 drivers/virtio/Kconfig                 |  11 +
 drivers/virtio/Makefile                |   1 +
 drivers/virtio/uvirtio.c               | 399 +++++++++++++++++++++++++
 include/linux/uvirtio.h                |  16 +
 include/uapi/linux/uvirtio.h           |  50 ++++
 samples/uvirtio/Makefile               |   9 +
 samples/uvirtio/uvirtio-vga.c          |  72 +++++
 9 files changed, 615 insertions(+)
 create mode 100644 Documentation/misc-devices/uvirtio.rst
 create mode 100644 drivers/virtio/uvirtio.c
 create mode 100644 include/linux/uvirtio.h
 create mode 100644 include/uapi/linux/uvirtio.h
 create mode 100644 samples/uvirtio/Makefile
 create mode 100644 samples/uvirtio/uvirtio-vga.c

diff --git a/Documentation/misc-devices/index.rst b/Documentation/misc-devices/index.rst
index c1dcd2628911..c87d4a8fd42b 100644
--- a/Documentation/misc-devices/index.rst
+++ b/Documentation/misc-devices/index.rst
@@ -21,4 +21,5 @@ fit into other categories.
    lis3lv02d
    max6875
    mic/index
+   uvirtio
    xilinx_sdfec
diff --git a/Documentation/misc-devices/uvirtio.rst b/Documentation/misc-devices/uvirtio.rst
new file mode 100644
index 000000000000..48ba81a00ce0
--- /dev/null
+++ b/Documentation/misc-devices/uvirtio.rst
@@ -0,0 +1,56 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+=================================
+User space created virtio devices
+=================================
+
+:Authors: - Lepton Wu <ytht.net@gmail.com> - 30 April 2020
+
+Introduction
+============
+
+Virtio devices are common in Linux VM guests. In most cases, there is a kernel driver in VM guest side and it communicates with VMM via virtio. To use specific virtio device, there should be correspond support in VMM side. But in some cases, there could be no support for the wanted virtio device at the VMM side or even no VMM at all. Uvirtio was developed for this. Instead of communicating with VMM, the uvirtio driver checks the virtio requests sent from virtio device driver and either "eats" it or replies with a "predefined" reply. This model works well for virtio-vga use case.
+
+The ioctl API
+=============
+
+To create a virtio device from user space, open /dev/uvirtio device first, then
+call ioctl with UV_DEV_EXPECT cmd to setup expected virtio request and the prepared reply. This must be done before calling UV_DEV_CREATE ioctl.
+
+Here is the example to setup that we are expecting one request in control queue which has a size of 100 request buffer and we are expecting that the offset 10 of the buffer should be "ABC" and the offset 20 of the buffer should be "DE"  and also we are expecting the size of reply buffer should be 30. If all of these matches, we set the offset of  10 of the reply buffer to "FGH"::
+
+	struct uvirtio_expect expect = { };
+
+	strcpy(expect.expects[0].vq_name, "control");
+
+	struct uvirtio_block *b = &expect.expects[0].blocks[0];
+	b->len = 100;
+	b->flags = VRING_DESC_F_NEXT;
+	b->data = "ABCDE";
+	b->rules[0].off = 10;
+	b->rules[0].len = 3;
+	b->rules[1].off = 20;
+	b->rules[1].len = 2;
+
+	b = &expect.expects[0].blocks[1];
+	b->len = 30;
+	b->flags = VRING_DESC_F_WRITE;
+	b->data = "FGH";
+	b->rules[0].off = 10;
+	b->rules[0].len = 3;
+
+After calling ioctl with UV_DEV_EXPECT cmd, call ioctl with UV_DEV_CREATE cmd to create virtio device::
+
+	struct uvirtio_setup setup;
+	struct virtio_gpu_config config = {.num_scanouts = 1 };
+
+	setup.features = 1ULL << VIRTIO_F_VERSION_1;
+	setup.config_addr = (__u64) &config;
+	setup.config_len = sizeof(config);
+	setup.id = VIRTIO_ID_GPU;
+	setup.flags = 1 << UVIRTIO_DEV_FLAG_STICK;
+	ret = ioctl(fd, UV_DEV_CREATE, &setup);
+
+By default, when the fd is closed, the created virtio device will be destroyed. By setting UVIRTIO_DEV_FLAG_STICK, the device won't be removed after the fd gets closed.
+
+See samples/uvirtio/uvirito-vga.c for a full example to create a virtio-vga device for user space.
diff --git a/drivers/virtio/Kconfig b/drivers/virtio/Kconfig
index 69a32dfc318a..4686df49cac5 100644
--- a/drivers/virtio/Kconfig
+++ b/drivers/virtio/Kconfig
@@ -109,4 +109,15 @@ config VIRTIO_MMIO_CMDLINE_DEVICES
 
 	 If unsure, say 'N'.
 
+config UVIRTIO
+	tristate "UVirtio driver"
+	select VIRTIO
+	help
+	 This driver supports creating virtio devices from userspace.
+
+	 This can be used to create virtio devices from user space without
+         supports from VMM. Check samples/uvirtio for examples.
+
+	 If unsure, say 'N'.
+
 endif # VIRTIO_MENU
diff --git a/drivers/virtio/Makefile b/drivers/virtio/Makefile
index 29a1386ecc03..558b2f890e8c 100644
--- a/drivers/virtio/Makefile
+++ b/drivers/virtio/Makefile
@@ -7,3 +7,4 @@ virtio_pci-$(CONFIG_VIRTIO_PCI_LEGACY) += virtio_pci_legacy.o
 obj-$(CONFIG_VIRTIO_BALLOON) += virtio_balloon.o
 obj-$(CONFIG_VIRTIO_INPUT) += virtio_input.o
 obj-$(CONFIG_VIRTIO_VDPA) += virtio_vdpa.o
+obj-$(CONFIG_UVIRTIO) += uvirtio.o
diff --git a/drivers/virtio/uvirtio.c b/drivers/virtio/uvirtio.c
new file mode 100644
index 000000000000..64cc9140de7a
--- /dev/null
+++ b/drivers/virtio/uvirtio.c
@@ -0,0 +1,399 @@
+// SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note
+/*
+ *  User level device support for virtio subsystem
+ */
+
+#include <linux/module.h>
+#include <linux/miscdevice.h>
+#include <linux/slab.h>
+#include <linux/uaccess.h>
+#include <linux/uvirtio.h>
+#include <linux/virtio.h>
+#include <linux/virtio_config.h>
+#include <linux/virtio_ring.h>
+#include <uapi/linux/uvirtio.h>
+
+#define UVIRTIO_MAX_EXPECT_DATA  (1UL << 20)
+
+struct uvirtio_device {
+	struct virtio_device vdev;
+	struct mutex mutex;
+	enum uvirtio_state state;
+	unsigned char virtio_status;
+	struct uvirtio_setup setup;
+	struct uvirtio_expect expect;
+	char *expect_data;
+};
+
+static struct miscdevice uvirtio_misc;
+
+static struct bus_type uvirtio_bus = {
+	.name = "",
+};
+
+static u64 uvirtio_get_features(struct virtio_device *dev)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	return udev->setup.features;
+}
+
+static int uvirtio_finalize_features(struct virtio_device *vdev)
+{
+	return 0;
+}
+
+static void uvirtio_get(struct virtio_device *dev, unsigned int offset,
+			void *buf, unsigned int len)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	if (WARN_ON(offset + len > udev->setup.config_len))
+		return;
+	memcpy(buf, (char *)udev->setup.config_addr + offset, len);
+}
+
+static u8 uvirtio_get_status(struct virtio_device *dev)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	return udev->virtio_status;
+}
+
+static void uvirtio_set_status(struct virtio_device *dev, u8 status)
+{
+	struct uvirtio_device *udev = container_of(dev, struct uvirtio_device,
+						   vdev);
+	if (WARN_ON(!status))
+		return;
+	udev->virtio_status = status;
+}
+
+static int find_match(int write, char *buf, unsigned int len,
+		      struct uvirtio_block *block, char *data)
+{
+	int i;
+	int off = 0;
+
+	for (i = 0; i < UVIRTIO_MAX_RULES; ++i) {
+		if (!block->rules[i].len)
+			break;
+		if (block->rules[i].off + block->rules[i].len > len)
+			return -1;
+		if (write) {
+			memcpy(buf + block->rules[i].off,
+			       data + block->data + off, block->rules[i].len);
+		} else {
+			if (memcmp(buf + block->rules[i].off,
+				   data + block->data + off,
+				   block->rules[i].len))
+				return -1;
+		}
+		off += block->rules[i].len;
+	}
+	return i ? write : 0;
+}
+
+static void process_vq(struct virtio_device *vdev, const char *name,
+		       struct vring_desc *desc, int idx)
+{
+	unsigned short flags;
+	char *buf;
+	unsigned int len;
+	struct uvirtio_device *udev = container_of(vdev, struct uvirtio_device,
+						   vdev);
+	struct uvirtio_expect *expect = &udev->expect;
+	int i, j, matched = 0;
+
+	for (i = 0; i < UVIRTIO_MAX_EXPECTS; ++i) {
+		if (!expect->expects[i].vq_name[0])
+			break;
+		if (strncmp(name, expect->expects[i].vq_name,
+			    sizeof(expect->expects[i].vq_name)))
+			continue;
+		for (j = 0; j < UVIRTIO_MAX_BLOCKS &&
+		     expect->expects[i].blocks[j].len; ++j) {
+			flags = virtio16_to_cpu(vdev, desc[idx].flags);
+			len = virtio32_to_cpu(vdev, desc[idx].len);
+			buf = __va(virtio64_to_cpu(vdev, desc[idx].addr));
+			if (expect->expects[i].blocks[j].len != len ||
+			    expect->expects[i].blocks[j].flags != flags)
+				break;
+			matched = find_match(flags & VRING_DESC_F_WRITE,
+					     buf, len,
+					     &expect->expects[i].blocks[j],
+					     udev->expect_data);
+			if (matched)
+				break;
+			if (!(flags & VRING_DESC_F_NEXT))
+				break;
+			idx = virtio16_to_cpu(vdev, desc[idx].next);
+		}
+		if (matched > 0)
+			break;
+	}
+}
+
+static bool uvirtio_notify(struct virtqueue *vq)
+{
+	struct vring *r = (struct vring *)(vq + 1);
+	int used_idx, avail_idx, id;
+
+	used_idx = virtio16_to_cpu(vq->vdev, r->used->idx);
+	avail_idx = virtio16_to_cpu(vq->vdev, r->avail->idx);
+	while (used_idx != avail_idx) {
+		id = used_idx & (r->num - 1);
+		process_vq(vq->vdev, vq->name, r->desc, r->avail->ring[id]);
+		r->used->ring[id].id = r->avail->ring[id];
+		used_idx++;
+	}
+	r->used->idx = r->avail->idx;
+	vq->callback(vq);
+	return true;
+}
+
+static int uvirtio_find_vqs(struct virtio_device *dev, unsigned int nvqs,
+			    struct virtqueue *vqs[],
+			    vq_callback_t *callbacks[],
+			    const char *const names[],
+			    const bool *ctx, struct irq_affinity *desc)
+{
+	int i, j;
+
+	for (i = 0; i < nvqs; ++i) {
+		vqs[i] = vring_create_virtqueue(i, 256, SMP_CACHE_BYTES, dev,
+						true, false, false,
+						uvirtio_notify, callbacks[i],
+						names[i]);
+		if (!vqs[i])
+			goto err;
+	}
+	return 0;
+err:
+	for (j = 0; j < i; ++j) {
+		vring_del_virtqueue(vqs[j]);
+		vqs[j] = NULL;
+	}
+	return -ENOMEM;
+}
+
+static void uvirtio_del_vqs(struct virtio_device *dev)
+{
+	struct virtqueue *vq, *n;
+
+	list_for_each_entry_safe(vq, n, &dev->vqs, list)
+		vring_del_virtqueue(vq);
+}
+
+static void uvirtio_reset(struct virtio_device *vdev)
+{
+}
+
+static const struct virtio_config_ops uvirtio_vq_ops = {
+	.get_features = uvirtio_get_features,
+	.finalize_features = uvirtio_finalize_features,
+	.get = uvirtio_get,
+	.get_status = uvirtio_get_status,
+	.set_status = uvirtio_set_status,
+	.reset = uvirtio_reset,
+	.find_vqs = uvirtio_find_vqs,
+	.del_vqs = uvirtio_del_vqs,
+};
+
+static int uvirtio_open(struct inode *inode, struct file *file)
+{
+	struct uvirtio_device *newdev;
+
+	newdev = kzalloc(sizeof(*newdev), GFP_KERNEL);
+	if (!newdev)
+		return -ENOMEM;
+
+	mutex_init(&newdev->mutex);
+	newdev->state = UVST_NEW_DEVICE;
+
+	file->private_data = newdev;
+	nonseekable_open(inode, file);
+
+	return 0;
+}
+
+static int uvirtio_release(struct inode *inode, struct file *file)
+{
+	struct uvirtio_device *udev = file->private_data;
+
+	if (udev->state == UVST_CREATED) {
+		if (udev->setup.flags & (1 << UVIRTIO_DEV_FLAG_STICK))
+			return 0;
+		unregister_virtio_device(&udev->vdev);
+	}
+	kfree((void *)udev->setup.config_addr);
+	kfree(udev->expect_data);
+	kfree(udev);
+	return 0;
+}
+
+static int load_data(struct uvirtio_block *b, char **ptr,
+		     unsigned int *off, unsigned int *len)
+{
+	int i, ret, total = 0;
+	void __user *p = (void __user *)b->data;
+	unsigned int need;
+	int realloc = 0;
+
+	for (i = 0; i < UVIRTIO_MAX_RULES; ++i) {
+		if (!b->rules[i].len)
+			break;
+		total += b->rules[i].len;
+	}
+	if (!total)
+		return 0;
+
+	need = (*off) + total;
+	while (need > (*len)) {
+		if (*len >= UVIRTIO_MAX_EXPECT_DATA)
+			return -ENOMEM;
+		*len = (*len) << 1;
+		realloc = 1;
+	}
+	if (realloc)
+		(*ptr) = krealloc(*ptr, *len, GFP_KERNEL);
+	if (!(*ptr))
+		return -ENOMEM;
+
+	ret = copy_from_user(*ptr + (*off), p, total);
+	if (ret)
+		return ret;
+	b->data = (*off);
+	(*off) += total;
+	return 0;
+}
+
+static int set_expect(struct uvirtio_device *udev, unsigned long arg)
+{
+	int i, j, ret;
+	struct uvirtio_block *b;
+	unsigned int off = 0, len = 1024;
+
+	ret = copy_from_user(&udev->expect, (void __user *)arg,
+			     sizeof(udev->expect));
+	if (ret)
+		return ret;
+
+	udev->expect_data = kmalloc(len, GFP_KERNEL);
+
+	if (!udev->expect_data)
+		return -ENOMEM;
+
+	for (i = 0; i < UVIRTIO_MAX_EXPECTS; ++i) {
+		if (!udev->expect.expects[i].vq_name[0])
+			break;
+		for (j = 0; j < UVIRTIO_MAX_BLOCKS; ++j) {
+			b = &udev->expect.expects[i].blocks[j];
+			if (!b->len)
+				break;
+			ret = load_data(b, &udev->expect_data, &off, &len);
+			if (ret)
+				goto err;
+		}
+	}
+	return 0;
+err:
+	kfree(udev->expect_data);
+	udev->expect_data = NULL;
+	return ret;
+}
+
+static int create_device(struct uvirtio_device *udev, unsigned long arg)
+{
+	int ret;
+	char *config = NULL;
+
+	ret = copy_from_user(&udev->setup, (void __user *)arg,
+			     sizeof(udev->setup));
+	if (ret)
+		return ret;
+	if (!udev->setup.config_len || !udev->setup.config_addr)
+		return -EINVAL;
+	if (udev->setup.config_len > UVIRTIO_MAX_CONFIG_LEN)
+		return -EINVAL;
+	if (udev->setup.flags & ~((1 << UVIRTIO_DEV_FLAG_MAX) - 1))
+		return -EINVAL;
+	config = kmalloc(udev->setup.config_len, GFP_KERNEL);
+	if (!config)
+		return -ENOMEM;
+	ret = copy_from_user(config, (void __user *)udev->setup.config_addr,
+			     udev->setup.config_len);
+	if (ret)
+		goto err;
+	udev->setup.config_addr = (uint64_t) config;
+
+	if (uvirtio_misc.this_device->bus == NULL)
+		uvirtio_misc.this_device->bus = &uvirtio_bus;
+
+	udev->vdev.dev.parent = uvirtio_misc.this_device;
+	udev->vdev.id.device = udev->setup.id;
+	udev->vdev.config = &uvirtio_vq_ops;
+	ret = register_virtio_device(&udev->vdev);
+	if (ret)
+		goto err;
+	udev->state = UVST_CREATED;
+	return 0;
+err:
+	kfree(config);
+	udev->setup.config_addr = 0;
+	return ret;
+}
+
+static long uvirtio_ioctl(struct file *file, unsigned int cmd,
+			  unsigned long arg)
+{
+	struct uvirtio_device *udev = file->private_data;
+	int ret;
+
+	ret = mutex_lock_interruptible(&udev->mutex);
+	if (ret)
+		return ret;
+
+	if (udev->state != UVST_NEW_DEVICE) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	switch (cmd) {
+	case UV_DEV_CREATE:
+		ret = create_device(udev, arg);
+		break;
+	case UV_DEV_EXPECT:
+		ret = set_expect(udev, arg);
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+out:
+	mutex_unlock(&udev->mutex);
+	return ret;
+}
+
+static const struct file_operations uvirtio_fops = {
+	.owner = THIS_MODULE,
+	.open = uvirtio_open,
+	.release = uvirtio_release,
+	.unlocked_ioctl = uvirtio_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl = uvirtio_ioctl,
+#endif
+	.llseek = no_llseek,
+};
+
+static struct miscdevice uvirtio_misc = {
+	.fops = &uvirtio_fops,
+	.minor = MISC_DYNAMIC_MINOR,
+	.name = "uvirtio",
+};
+
+module_misc_device(uvirtio_misc);
+
+MODULE_VERSION("0.0.1");
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("User level driver support for virtio subsystem");
diff --git a/include/linux/uvirtio.h b/include/linux/uvirtio.h
new file mode 100644
index 000000000000..ae3fd6dfbe58
--- /dev/null
+++ b/include/linux/uvirtio.h
@@ -0,0 +1,16 @@
+/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
+/*
+ * User level device support for virtio subsystem
+ *
+ * Check samples/uvirtio for examples.
+ *
+ * Based on uinput.c by Aristeu Sergio Rozanski Filho
+ */
+#ifndef __UVIRTIO_H_
+#define __UVIRTIO_H_
+
+#include <uapi/linux/uvirtio.h>
+
+enum uvirtio_state { UVST_NEW_DEVICE, UVST_CREATED };
+
+#endif  /* __UVIRTIO_H_ */
diff --git a/include/uapi/linux/uvirtio.h b/include/uapi/linux/uvirtio.h
new file mode 100644
index 000000000000..a17f923d73a8
--- /dev/null
+++ b/include/uapi/linux/uvirtio.h
@@ -0,0 +1,50 @@
+/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
+#ifndef _UAPI_UVIRTIO_H_
+#define _UAPI_UVIRTIO_H_
+
+#include <linux/types.h>
+
+/* ioctl */
+
+#define UVIRTIO_MAX_CONFIG_LEN 1024
+#define UVIRTIO_MAX_RULES 8
+#define UVIRTIO_MAX_BLOCKS 8
+#define UVIRTIO_MAX_EXPECTS 8
+#define UVIRTIO_MAX_VQ_NAME 64
+
+enum {
+	UVIRTIO_DEV_FLAG_STICK = 0,
+	UVIRTIO_DEV_FLAG_MAX
+};
+
+struct uvirtio_setup {
+	__u64 features;
+	__u64 config_addr;
+	__u16 config_len;
+	__u16 flags;
+	__u32 id;
+};
+
+struct uvirtio_block {
+	__u32 len;
+	__u16 flags;
+	__u16 unused;
+	__u64 data;
+	struct {
+		__u16 off;
+		__u16 len;
+	} rules[UVIRTIO_MAX_RULES];
+};
+
+struct uvirtio_expect {
+	struct {
+		char vq_name[UVIRTIO_MAX_VQ_NAME];
+		struct uvirtio_block blocks[UVIRTIO_MAX_BLOCKS];
+	} expects[UVIRTIO_MAX_EXPECTS];
+};
+
+#define UVIRTIO_IOCTL_BASE	'V'
+#define UV_DEV_CREATE	_IOW(UVIRTIO_IOCTL_BASE, 1, struct uvirtio_setup)
+#define UV_DEV_EXPECT	_IOW(UVIRTIO_IOCTL_BASE, 2, struct uvirtio_expect)
+
+#endif /* _UAPI_UVIRTIO_H_ */
diff --git a/samples/uvirtio/Makefile b/samples/uvirtio/Makefile
new file mode 100644
index 000000000000..ee830986eda2
--- /dev/null
+++ b/samples/uvirtio/Makefile
@@ -0,0 +1,9 @@
+uvirtio-vga: uvirtio-vga.c uapi/linux/uvirtio.h linux/uvirtio.h
+	$(CC) -o $@ $^ -I.
+uapi/linux/uvirtio.h:
+	mkdir -p uapi/linux && ln ../../../../include/uapi/linux/uvirtio.h uapi/linux -sf
+linux/uvirtio.h:
+	mkdir -p linux && ln ../../../include/linux/uvirtio.h linux -sf
+.PHONY: clean
+clean:
+	rm -f uvirtio-vga
diff --git a/samples/uvirtio/uvirtio-vga.c b/samples/uvirtio/uvirtio-vga.c
new file mode 100644
index 000000000000..fae79a6f4968
--- /dev/null
+++ b/samples/uvirtio/uvirtio-vga.c
@@ -0,0 +1,72 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <stdio.h>
+#include <stddef.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/ioctl.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <linux/uvirtio.h>
+#include <linux/virtio_config.h>
+#include <linux/virtio_ids.h>
+#include <linux/virtio_gpu.h>
+#include <linux/virtio_ring.h>
+
+void error(char *str)
+{
+	fprintf(stderr, "%s\n", str);
+	exit(-1);
+}
+
+int main(void)
+{
+	int ret;
+	int fd = open("/dev/uvirtio", O_RDWR);
+
+	if (fd < 0)
+		error("open");
+
+	struct uvirtio_expect expect = { };
+
+	strcpy(expect.expects[0].vq_name, "control");
+
+	struct uvirtio_block *b = &expect.expects[0].blocks[0];
+	struct virtio_gpu_ctrl_hdr cmd;
+
+	cmd.type = VIRTIO_GPU_CMD_GET_DISPLAY_INFO;
+	b->len = sizeof(cmd);
+	b->flags = VRING_DESC_F_NEXT;
+	b->data = (__u64) &cmd.type;
+	b->rules[0].off = offsetof(struct virtio_gpu_ctrl_hdr, type);
+	b->rules[0].len = sizeof(cmd.type);
+
+	struct virtio_gpu_resp_display_info info = { };
+
+	info.pmodes[0].r.width = 1024;
+	info.pmodes[0].r.height = 768;
+	info.pmodes[0].enabled = 1;
+	b = &expect.expects[0].blocks[1];
+	b->len = sizeof(info);
+	b->flags = VRING_DESC_F_WRITE;
+	b->data = (__u64) &info.pmodes[0];
+	b->rules[0].off =
+	    offsetof(struct virtio_gpu_resp_display_info, pmodes[0]);
+	b->rules[0].len = sizeof(info.pmodes[0]);
+	ret = ioctl(fd, UV_DEV_EXPECT, &expect);
+	if (ret < 0)
+		error("ioctl UV_DEV_EXPECT");
+
+	struct uvirtio_setup setup;
+	struct virtio_gpu_config config = {.num_scanouts = 1 };
+
+	setup.features = 1ULL << VIRTIO_F_VERSION_1;
+	setup.config_addr = (__u64) &config;
+	setup.config_len = sizeof(config);
+	setup.id = VIRTIO_ID_GPU;
+	setup.flags = 1 << UVIRTIO_DEV_FLAG_STICK;
+	ret = ioctl(fd, UV_DEV_CREATE, &setup);
+	if (ret < 0)
+		error("ioctl UV_DEV_CREATE");
+}
-- 
2.26.2.526.g744177e7f7-goog

Re: [PATCH 0/1] Add uvirtio for testing

[lepton]

On Thu, Apr 30, 2020 at 12:51 AM Gerd Hoffmann <kraxel@redhat.com> wrote:
>
> On Wed, Apr 29, 2020 at 08:59:18PM -0700, lepton wrote:
> > On Wed, Apr 29, 2020 at 4:58 AM Gerd Hoffmann <kraxel@redhat.com> wrote:
> > >
> > > > 3) Need to be verbose on how the vring processing work in the commit log of
> > > > patch 1
> > >
> > > Ecven better a file documenting the interface somewhere in
> > > Documentation/
> > I put a uvirtio-vga.c under samples/uvirtio and hope this can serve
> > the purpose of the document since currently that's the only tested use
> > case.
>
> Sample code is good, but *commented* sample code would be better.
>
> > Maybe have a document later if this really get more use cases?
>
> Well, having good documentation right from the start would help getting
> more use cases ;)
Now I add document in V3.

Thanks. Regards
>
> take care,
>   Gerd
>

Re: [PATCH 0/1] Add uvirtio for testing

[Jason Wang]

On 2020/4/30 上午11:56, lepton wrote:
> On Wed, Apr 29, 2020 at 2:57 AM Jason Wang <jasowang@redhat.com> wrote:
>>
>> On 2020/4/29 上午4:47, Lepton Wu wrote:
>>> This is a way to create virtio based devices from user space. This is the
>>> background for this patch:
>>>
>>> We have some images works fine under qemu, we'd like to also run the same image
>>> on Google Cloud. Currently Google Cloud doesn't support virtio-vga. I had a
>>> patch to create a virtio-vga from kernel directly:
>>> https://www.spinics.net/lists/dri-devel/msg248573.html
>>>
>>> Then I got feedback from Gerd that maybe it's better to change that to something
>>> like uvirtio. Since I really don't have other use cases for now, I just implemented the minimal stuff which work for my use case.
>>
>> Interesting, several questions:
>>
>> 1) Are you aware of virtio vhost-user driver done by UM guys?
>> (arch/um/drivers/virtio_uml.c) The memory part is tricky but overall
>> both of you have similar target.
> Thanks for reminding me, I was not aware of it. The use case looks a
> little different: they are trying create virtio devices for user mode
> linux and it communicated with "host" side. My driver doesn't depends
> on any HOST/VMM side stuff.


Yes, but as I said. Except for the memory table part, the rest could be 
probably reused?

The only difference to me is that, you develop your own uABI while 
virtio_uml uses vhost-user.


> Basically we can use it to create virtio
> device from guest itself. Or even create virtio device on bare metal.
>> 2) Patch 1 said it's userspace virtio driver, which I think it is
>> actually "userspace virtio device"
> Updated in version 2 of this patch.
>> 3) Need to be verbose on how the vring processing work in the commit log
>> of patch 1
> Updated.
>> 4) I'm curious which testing you want to accomplish through this new
>> transport, I guess you want to test a specific virtio driver?
> Here is the whole story: we want to test our custom linux image. In
> the past, we just test our custom linux image with qemu (and virtio
> vga), and run qemu session on Google Cloud. As you can see, this is
> nested virtualization and performance hurts. And more, we have another
> vm inside our custom linux image. So we want to remove the qemu layer,
> just run our custom linux image on Google Cloud directly. Then we need
> some kind of VGA.  So a "dummy" virtio vga looks a good fit.


Then it's better to mention this in the cover letter.


>> 5) You mentioned that you may want to develop communication between
>> kernel and userspace, any more details on that?
> Currently, we don't have such use case. But maybe others can
> furthermore to extend uvirtio for this. For example, user space can
> use read/write to actually exchange data with kernel. Then that could
> be used for simulate more complex use case.


OK.

Thanks


>> Thanks
>>
>>
>>> Lepton Wu (1):
>>>     virtio: Add uvirtio driver
>>>
>>>    drivers/virtio/Kconfig        |   8 +
>>>    drivers/virtio/Makefile       |   1 +
>>>    drivers/virtio/uvirtio.c      | 405 ++++++++++++++++++++++++++++++++++
>>>    include/linux/uvirtio.h       |   8 +
>>>    include/uapi/linux/uvirtio.h  |  69 ++++++
>>>    samples/uvirtio/Makefile      |   9 +
>>>    samples/uvirtio/uvirtio-vga.c |  63 ++++++
>>>    7 files changed, 563 insertions(+)
>>>    create mode 100644 drivers/virtio/uvirtio.c
>>>    create mode 100644 include/linux/uvirtio.h
>>>    create mode 100644 include/uapi/linux/uvirtio.h
>>>    create mode 100644 samples/uvirtio/Makefile
>>>    create mode 100644 samples/uvirtio/uvirtio-vga.c
>>>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization